1. Field of the Invention
The present invention relates to a technique for reproducing an image based on an image signal obtained by scanning each line, utilized in an electro-photographic printer such as a laser printer, or an LED printer with an LED array head, a CRT, or the like. More particularly, the present invention relates to an image processing method and an image processing apparatus for smoothing jagged edges of characters, etc., and for stably reproducing thin lines and isolated dots, so as to achieve an image reproduction of high quality and to achieve an ideal tone characteristic by correcting the tone characteristic of digital halftone images.
2. Description of the Related Art
There are two conventional methods for improving an image quality by smoothing edges of images reproduced by a printer or a CRT. In the first method, image data of higher resolution is generated, so that the resolution of the image reproduction apparatus is increased.
However, this method costs much because it necessitates a memory with larger capacity for storing the generated image data. Moreover, it costs more because the exposure system should have a higher accuracy for the image reproduction apparatus with the increased resolution.
The second method is disclosed in U.S. Pat. No. 4,847,641 to Tung et al. entitled "PIECE-WISE PRINT IMAGE ENHANCEMENT FOR DOT MATRIX PRINTERS", published on Jul. 11, 1989.
An image processing apparatus utilizing the
second method will be described below with reference to FIGS. 33, 34A to 34D, and 35.
FIG. 33 shows a configuration of a conventional image processing apparatus for smoothing edges. A bit map image signal 100 obtained by raster scan is input into a pixel window scanning circuit 101. The pixel window scanning circuit 101 includes a line buffer memory for affecting a line delay and a shift register for affecting a pixel delay. The pixel window scanning circuit 101 outputs data 102 of M.times.N pixels. In the M.times.N pixels, the center pixel is a pixel to be processed. The data 102 of M.times.N pixels are input into a pattern matching and corrected image signal generating circuit 103 in which a plurality of predetermined patterns are compared with the data 102. If the data 102 match any one of the patterns, a pattern match signal 106 is made to be active. When the pattern match signal 106 is active, the pattern matching and corrected image signal generating circuit 103 outputs corrected center pixel data 105.
The corrected center pixel data 105 and non-corrected center pixel data 104 are input into a selector 107. When the pattern match signal 106 is active, the selector 107 selects the corrected center pixel data 105. The output 108 of the selector 107 is used as a modulation signal for a semiconductor laser in a laser printer.
Next, how the edges are smoothed in the above conventional image processing apparatus will be described below.
FIG. 34A shows input image data. When the edge smoothing is not to be performed, the laser modulation signal for the input image data is shown by solid lines in FIG. 34B. Even if it is considered that the spot shape of a laser beam is circular, the reproduced image has jagged edges as shown in FIG. 34C.
When the edge smoothing is to be performed, pixels corresponding to portions A, B, C, D, E, and F in FIG. 34A are detected by pattern matching, so that the laser modulation signal is corrected. For example, a template pattern shown in FIG. 35 is compared with data of 9.times.9 pixels in which the center pixel is a pixel to be processed. In FIG. 35, a dot meshing pixel indicates a pixel which is not cared for during the comparison, a hatched pixel indicates an exposed pixel, and a white pixel indicates a non-exposed pixel. If the comparison results in matching, the center pixel data in which the pixel is not exposed to laser light is replaced by data in which the right half of the pixel is exposed to laser light, as is shown in FIG. 36. The template pattern of FIG. 35 matches the data for processing the pixel E in FIG. 34A. A plurality of types of template patterns such as that shown in FIG. 35 are prepared. When the data to be processed matches any one of the template patterns, the center pixel data is replaced so as to smooth the jagged edges. For example, the center pixel data is replaced so as to have a corrected laser modulation signal shown by dotted lines in FIG. 34B. As is shown in FIG. 34D, the reproduced image based on the corrected laser signal has less lagged edges. In this way, an image with higher resolution can be obtained.
However, in cases where an isolated dot consisting of 1 pixel or a thin line of 1-pixel width is to be reproduced by using the above edge smoothing method, since the shape of the exposure beam spot has a broad profile such as Gaussian distribution, the reproduced thin line may not be clear or may be thinner, and the isolated dot cannot be accurately reproduced. In order to reproduce thin lines and isolated dots, it is necessary to set a large value for the exposure energy per one pixel. However, with such a method, there were problems such as the fattening of font images and the smothering out of font images.
Moreover, according to the conventional edge smoothing process, the exposure energy density per one pixel for a pixel to be exposed cannot be made higher than in the case where the smoothing process is not performed, so that there arises a problem in that a sufficient edge smoothing effect cannot be obtained.
If the input image data is a digital halftone image, the conventional edge smoothing process has such drawbacks that a pseudo-contour and a moire may occur and that the tone characteristic may be degraded. In a conventional method for eliminating the above drawbacks of the edge smoothing process for the digital halftone image, the edge smoothing process is manually prohibited or the edge smoothing effect is manually attenuated. However, such a method requires manipulation by an operator, and in cases where font images and digital halftone images co-exist in the input image data, this method is not at all applicable.
Furthermore, the digital halftone image digitally represents the halftone in accordance with the ratio of black dots, so that the image density (or reflectance) reproduced by the printer should have a linear relationship to the black dot density. However, in the usual printer, the relationship between the black dot density and the reproduced image density is non-linear, so that it is difficult to accurately reproduce the halftone image with high quality.